Global AI Race Intensifies: China Closes Gap with US as Competition Heats Up

Global AI Landscape Becomes Increasingly Competitive

The artificial intelligence (AI) race has intensified significantly, with the performance gap between top models narrowing and new players emerging on the global stage. According to the 2025 Artificial Intelligence Index Report released by Stanford University's Institute for Human Centered AI, the AI frontier is "increasingly competitive -- and increasingly crowded" 1

1

.

US Lead Narrows as China Catches Up

While the United States remains the top producer of notable AI models, China is rapidly closing the performance gap. In 2024, the US released 40 notable models compared to China's 15 1

1

. However, Chinese models have reached near parity with US models on key benchmarks such as the Massive Multitask Language Understanding (MMLU) test. The performance difference between leading US and Chinese models on the MMLU has shrunk from nearly 20 percentage points in 2023 to just 0.3 percentage points by the end of 2024 2

2

.

Rise of Smaller, More Efficient AI Models

The report highlights 2024 as a breakthrough year for smaller AI models. Thanks to improved algorithms, modern compact models can now match the performance of models 100 times larger from just two years ago 1

1

. This trend towards efficiency is evident in the dramatic reduction in the number of parameters required to achieve high scores on benchmark tests 3

3

.

Emergence of Open-Weight Models

The AI field has seen a surprising surge in the performance of 'open-weight' models, such as Meta's LLaMa and China's DeepSeek. These models allow users to view the parameters learned during training, promoting transparency and accessibility. The performance gap between open-weight and closed-weight models has narrowed significantly, from 8% in early 2024 to just 1.7% in early 2025 4

4

.

Shift from Academia to Industry

The report notes a significant shift in AI model development from academia to industry. Before 2006, industry produced fewer than 20% of notable AI models. However, this figure rose to 60% in 2023 and nearly 90% in 2024 1

1

. This trend reflects the growing commercial importance and application of AI technologies.

Global Expansion of AI Development

While the US and China dominate the AI landscape, other regions are joining the race. Notable AI launches have come from the Middle East, Latin America, and Southeast Asia 5

5

. Europe, particularly France with Mistral AI, is also making strides in the field 5

5

.

Challenges in AI Evaluation

As AI models continue to improve, traditional benchmarks are becoming saturated, making it increasingly difficult to accurately compare new models. The industry is responding by developing more sophisticated evaluation methods, such as Humanity's Last Exam and Arena-Hard-Auto 4

4

.

Future Outlook

The report suggests that the AI race is likely to remain highly competitive, with no clear leader emerging. As smaller teams and companies demonstrate their ability to create innovative AI solutions, the field is expected to become even more diverse and dynamic. This increased competition is seen as a positive development, fostering innovation and preventing dominance by a few large companies 1

1

.